v0.1.2: improved vars, inspect and docs

agentnb 0.1.2

0.1.2 focuses on agent ergonomics: clearer execution semantics, lower-noise inspection, better reload/error guidance, and a much faster test suite.

Highlights

• Improved CLI usability for agents

  • Root flags like --agent and --json now work before or after subcommands.
  • Help text and guidance now make the intended workflow more explicit.
  • SESSION_BUSY errors are more actionable and clearly explain serial session use.

• Better compact outputs

  • vars is cleaner and more useful for exploratory sessions.
  • inspect gives structural previews for common containers instead of mostly generic object noise.
  • history is more semantic and less code-heavy by default.
  • JSON error payloads now use compact, ANSI-free tracebacks.

• Better debugging ergonomics

  • reload now reports rebound names and warns more clearly about stale aliases and instances.
  • Exec and reset failures are surfaced as real command errors instead of nested “ok” payloads.

• Better discoverability

  • README.md, SKILL.md, CLI help, and command suggestions now explain:
    • notebook-style exec behavior
    • final expression vs print(...)
    • one-command-at-a-time session usage
    • vars --recent and vars --match for namespace hygiene

• Faster tests

  • CLI tests were split into cheaper command-layer coverage where possible.
  • Runtime and ops integration tests now reuse warm kernels.
  • Full suite runtime dropped substantially, from roughly 88s to about 20s.

Notable additions

• vars --recent N
• vars --match TEXT
• improved sqlite row previews
• semantic history labels for failed execs

Validation

• uv run pytest
• uv run ruff check src tests
• uv run ruff format --check src tests